INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ems

-0.74

 vest

-0.71

ridges

-0.70

enary

-0.67

 balance

-0.66

rontal

-0.66

 chances

-0.65

rack

-0.64

 crest

-0.64

isites

-0.64

POSITIVE LOGITS

Jess

0.78

NYSE

0.74

EEK

0.70

 Jessie

0.70

Â±

0.69

UGC

0.68

Linux

0.67

PokÃ©mon

0.67

Community

0.66

meat

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.